Multiple Small RNAs Alter Membrane Lipid Composition via Post-Transcriptional Regulation of Cyclopropane Fatty Acid Synthase

Microbial membranes are the first line of defense against environmental stress as well as the site of many metabolic processes. As such, maintenance of membrane integrity and homeostasis is key to cell survival. Altering membrane protein and lipid composition is an important strategy for maintaining membrane integrity in response to many environmental stresses. There are now numerous examples of small RNA (sRNA)-mediated regulation of membrane protein production, but less is known about how sRNAs regulate the types and relative proportions of different fatty acids in the membrane. The only sRNA known to regulate membrane fatty acid composition is RydC, which stabilizes cfa mRNA, encoding cyclopropane fatty acid (CFA) synthase, resulting in increased production of the synthase and higher levels of cyclopropane fatty acids (CFAs) in the cell membrane. Here, we report that three additional sRNAs, ArrS, CpxQ, and GadF also alter cfa translation and thus the amount of CFAs in cell membranes. RydC, ArrS, and GadF bind at sites that overlap a known RNase E cleavage site in the cfa mRNA 5′-untranslated region (UTR), resulting in increased cfa mRNA stability and translation. In contrast, CpxQ binds to a different site in the cfa mRNA 5′-UTR, and acts to reduce cfa translation. The physiological role of CFAs in membrane lipids is poorly understood, but CFAs have been shown to promote bacterial resistance to acid stress. We show that cfa translation increases in an sRNA-dependent manner when cells are subjected to mild acid stress. RydC is necessary for an increase in cfa translation at pH 5.0 and a rydC mutant is more sensitive to acid shock than the rydC+ strain. Alteration of membrane lipid composition is a key mechanism for bacterial responses to many environmental stresses, including acid stress. This work suggests an important role for sRNAs in these responses through their regulation of cfa mRNA.

of CFAs in phospholipids remains unclear. However, the conversion of UFAs to CFAs is tuberculosis leaves the bacteria unable to establish a persistent infection (4). One hypothesis is 70 that the formation of CFAs may reduce membrane fluidity and permeability, thereby preventing 71 undesirable molecules from entering the cell (3). Chang and Cronan identified the first 72 phenotype for CFAs, which is protection from acid shock (7,8). Escherichia coli cfa mutants 73 were unable to survive a rapid shift from pH 7 to pH 3. CFAs in membranes have been shown to 74 protect bacteria from acid shock because the conversion of UFA to CFA decreases permeability 75 of the membrane to protons (9). CFAs are thought to be important for pathogenesis because an 76 acid shock such as rapid shift from pH 7 to pH 3 occurs as an ingested pathogen goes through 77 the stomach (2). In fact, it has been noted that pathogenic E. coli strains contain more CFAs 78 and are more resistant to acid than nonpathogenic E. coli strains (8).

79
Production of CFAs is regulated in part by control of cfa transcription. In E. coli and 80 Salmonella, the cfa gene has two promoters (10, 11). The distal promoter is σ 70 -dependent and

87
Recently, cfa mRNA was also shown to be post-transcriptionally controlled by the sRNA 88 RydC (12). RydC is 64-nucleotides (nt) long and binds to the RNA chaperone Hfq (13   142 suggested that ArrS, GadF and CpxQ would base pair in the same region of cfa mRNA as RydC 143 (Fig. 1A, "RydC BS"), so we hypothesized that these sRNAs would alter the translation of the 144 P BAD -cfa'-'lacZ-Long but not the P BAD -cfa'-'lacZ-Short. When the individual sRNAs were 145 ectopically expressed from a plasmid, all sRNAs affected P BAD -cfa'-'lacZ-Long activity (Fig. 1B) and had no effect on P BAD -cfa'-'lacZ-Short activity (Fig. 1C). RydC, ArrS, and GadF activated 147 the long fusion 18-, 10-, and 1.3-fold, respectively, while CpxQ repressed the long fusion ~3-   Table S1). The FA composition of all strains was similar except for CFA 161 and UFA content (Table S1). In strains carrying P lac -cfa, there were reduced levels of 16:1 and       in the ΔcpxQ mutant suggests that at neutral pH, CpxQ is produced at sufficient levels to 218 repress cfa. The ΔcpxQ mutation had no effect on P BAD -cfa'-'lacZ-Long at pH 5, where levels 219 were similar to wild-type. The ΔrydC mutant had lower cfa'-'lacZ activity at both pH 7 and pH 5 220 compared to wild-type, suggesting that RydC may be produced at sufficient levels at both 221 neutral and acidic pH to have an activating effect on cfa translation under both conditions. We

234
Compared with the ΔrydC parent, the ΔcpxQ ΔrydC strain had slightly higher P BAD -cfa'-

235
'lacZ-Long activity at pH 7 (Fig. 5B). In contrast, at pH 5, the ΔrydC parent and ΔcpxQ ΔrydC 236 strain had similar levels of P BAD -cfa'-'lacZ-Long activity (Fig. 5B). Altogether, the data are  The σ s -dependent cfa promoter was previously implicated in increased CFA levels when 243 cells were grown at pH 5 compared to pH 7 (7). P BAD -cfa'-'lacZ-Long contains region 244 encompassing the σ s dependent promoter, so it was possible that RpoS could be regulating cfa 245 transcription in the context of the P BAD -cfa'-'lacZ-Long during mild acid stress. To determine if 246 RpoS impacts regulation of cfa under our conditions, we measured P BAD -cfa'-'lacZ-Long activity 247 at pH 5 and 7 in a ΔrpoS background (Fig. 6A). Deletion of rpoS had no effect on P BAD -cfa'-

248
'lacZ-Long activity at pH 5 or 7 and we observed the same increase in activity in response to pH 249 5 in both wild-type and ΔrpoS backgrounds, indicating that RpoS does not impact the observed 250 activity of the P BAD -cfa'-'lacZ-Long fusion.

295
Long, suggesting that the -212 to -190 region of the 5¢ UTR cfa mRNA may play a role in cfa 296 mRNA structure and stability. Based on these data, we hypothesize that ArrS and GadF 297 regulate cfa by the same mechanism as RydC, stabilizing the cfa mRNA transcript by impairing 298 RNAse E-mediated decay (12). CpxQ-mediated repression must occur via a different 299 mechanism that has yet to be determined.

315
5A,B). Acid stress phenotypes were also apparent for strains ectopically expressing ArrS (Fig.   316 3), which may be due to both ArrS-mediated activation of cfa and gadE.

317
Previous studies determined that the RydC-cfa mRNA base pairing prevents RNase E-318 mediated decay and stabilizes the mRNA to allow increased translation (12). We hypothesize, 319 because ArrS and GadF base pair with the same region of cfa mRNA as RydC, that ArrS and 320 GadF would regulate cfa mRNA stability via a similar mechanism. However, CpxQ must be 321 repressing cfa translation by a different mechanism. CpxQ has been shown to repress other 322 mRNA targets using one of two conserved seed regions (17). One of these seed regions    stress, a condition where CFA synthase is known to play a role (7, 8). We determined that cfa 341 translation was higher when cells were exposed to pH 5 compared to pH 7. When rydC was 342 deleted, cfa translation could not reach maximum wild-type levels at pH 5 and DrydC mutants 343 were more susceptible to acid shock compared to wild-type, giving us the first phenotype

365
CpxQ is the only sRNA of the four we have characterized that represses cfa translation.

366
Deletion of cpxQ resulted in higher cfa translation at pH 7 compared to wild-type, but did not

430
Gibson product was transformed into XL10 competent cells and the plasmid was purified. The 431 mutated cfa 5'UTR fragment of interest was PCR amplified from this plasmid using primers 432 containing 5' homologies to P BAD and lacZ (Table S3)

440
England Biolabs) to produce the plasmids.

441
P LlacO -arrS1, P LlacO -gadF1, and P LlacO -rydC6 were created using QuikChange mutagenesis 442 (Agilent Technologies) using primers in Table S3.     Transcription from the proximal promoter is controlled by σ s and produces a shorter cfa transcript with a 34-nt 5¢ UTR. (B-C Top) Two cfa translational fusions to lacZ (controlled by P BAD promoter) were constructed. The P BAD -cfa'-'lacZ-Long fusion (B) is from the distal σ 70 dependent promoter which contains a 212-nt 5¢ UTR that includes the RydC binding site (indicated by bar labeled "RydC BS") and the sites predicted for the ArrS, CpxQ, and GadF. The P BAD -cfa'-'lacZ Short fusion (C) contains only proximal the σ s dependent promoter and consequently not the predicted sRNA binding sites. (B-C Bottom) P BAD -cfa'-'lacZ-Long (B) and P BAD -cfa'-'lacZ-Short (C) carrying an empty, P lac -arrS, P lac -cpxQ, P lac -gadF, or P lac -rydC plasmid were grown in TB medium with 0.002% L-arabinose to early exponential phase then 0.1 mM IPTG was added to induce sRNA expression. Samples were harvested 60 minutes later and assayed for β-galactosidase activity of the reporter fusion. Error bars represent standard deviation for three biological replicates.  . Survival during an acid challenge. Strains were grown overnight in LB then subcultured into LB at pH 7.0 and grown at 37°C with shaking. Plasmids were induced with 0.1mM IPTG. When strains reached an OD 600 of 0.2, cultures were diluted 10x into LB pH 3.0 and incubated at 37°C with no shaking for 60 minutes. Survival was determined by the ratio of CFUs on the LB plates after acid shock to CFUs on the LB plate before acid shock. Error bars represent average ± standard deviation of 3 technical replicates. Figure 4: cfa translation is induced at acidic pH. Cells carrying (A) P BAD -cfa'-'lacZ-Long, (B) P BAD -cfa'-'lacZ-Short, or (C) P BAD -cfa'-'lacZ-DsRNABS were grown in TB medium pH 7.0 with 0.002% L-arabinose to early exponential phase then cells were subcultured into TB medium with 0.002% L-arabinose at either pH 7.0 (pH 7.0) or pH 5.0 (pH 5.0). Samples were harvested 120 minutes later and assayed for βgalactosidase activity of the reporter fusion. Data was analyzed as described in Figure 1.   Figure 4. (B) P BAD -cfa'-'lacZ-Long in a WT or ΔrpoS background carrying an empty, P lac -arrS, P lac -cpxQ, P lac -gadF, or P lac -rydC plasmid were grown as described in Figure 1.  Figure 7: Deletion of the putative sRNA binding site removes regulation of cfa translation by ArrS, CpxQ, GadF, and RydC (A) 5¢ UTR of cfa gene. Arrows mark transcriptional start sites, sRNA binding sites are underlined, and -79 nt is marked by a red arrow. (B) A cfa translational fusion to lacZ (P BADcfa'-'lacZ-DsRNABS) that begins immediately downstream of the predicted sRNA binding site was constructed. This fusion contains the proximal σ s dependent promoter and 79-nt upstream of this promoter. (B) P BAD -cfa'-'lacZ-DsRNABS cells carrying an empty, P lac -arrS, P lac -cpxQ, P lac -gadF, or P lac -rydC plasmid were grown as described in Figure 1. Data was analyzed as described in Figure 1. ) and WT sRNAs were tested for activation of WT P BAD -cfa'-'lacZ-Long as described in Fig. 1B. (E) Three point mutations (U103A, C102G, G101C) were made in P BAD -cfa'-'lacZ-Long (called P BAD -cfa'-'lacZ-LongAGC). Mutant sRNAs (ArrS1, GadF1, RydC6) and WT sRNAs were tested for activation of P BAD -cfa'-'lacZ-LongAGC as described in Fig. 1B.  Figure 9: CpxQ cannot repress cfa translation when putative CpxQ binding site is deleted. (A) The last 20 nt of the -212-5¢ UTR (-210 to -190) which includes the putative CpxQ binding site, was deleted in P BAD -cfa'-'lacZ-Long fusion (called P BAD -cfa'-'lacZ-DCpxQBS). (B) WT P BAD -cfa'-'lacZ-Long or P BAD -cfa'-'lacZ-DCpxQBS carrying an empty or P lac -cpxQ plasmid were grown as described in Figure 1. Data was analyzed as described in Figure 1. Table S1: Expression of sRNAs can alter fatty acid composition: Relative qualification of fatty acids in E. coli in response to ectopic expression of arrS, cfa, gadF, cpxQ, rydC or vector control. Fatty acids are presented as a percent of total identified fatty acids. Error represents average ± standard deviation, n=3 or *n=2. ND: not detected Figure S1: Deletion of sRNAs do not affect P BAD -cfa'-'lacZ -Short activity. (A) Cells carrying P BAD -cfa'-'lacZ -Short, as described in Figure 1C, in either a WT background or a background where one sRNA is deleted were grown as described in Figure 4. (B) Cells carrying P BAD -cfa'-'lacZ-Short (in either a WT background or a background where rydC and one other sRNA are deleted) were grown as described in Figure 4. ΔrydC single mutant is included for reference.